Analysis of spatio-temporal dynamics of urban sprawl and growth pattern using geospatial technologies and landscape metrics in Bahir Dar, Northwest Ethiopia

1. Introduction

In the current increasingly and interconnected world, more than half of the global population presently lives in urbanized areas and this level of urbanization is predicted to raise to 66% in the near 2050. In this figure, 2.5 billion will be added to the global population, almost 90% of raise stuffed in Developing Countries (UN, 2014). Though urbanization has started lately in developing countries, the speed of urban growth is very high than developed countries. These countries are least urbanized and characterized by unplanned urbanization which is manifested by misbalance between level of urbanization and their economy as well as industrial development (Sven et al., 2008). Various previously conducted studies ( e.g. Tegenu, 2010; Pravitasari, 2015; Bapari, 2016; Bodo, 2019; Noor and Rosni, 2013) indicate that rapid population growth and socio-economic development are major forces intensifying urban sprawl. Urban expansion is considered as an important phenomenon hence it offers various opportunities such as employment, technologies and inventions, production, goods and services (Güneralp et al., 2017, Bryan et al., 2019, Harris SELOD, 2017, Kötter, 2015, Mosel et al., 2016, UN, 2018). These opportunities further enhanced rural-urban migration to get accesses that are considered as different from rural areas (Cohen, 2006).

Though urban areas occupied small fraction of land as compared to other land use types, rapid urban growth in different parts of the world is resulting fast transformation of the urban land use classes such as agricultural land, forest, wetland and water bodies. As previous studies conducted on this area indicated, urban expansion; which in turn results urban sprawl in different parts of developing countries, led to fragmentation and diminish in cropland in the surrounding urban landscape (Bhat et al., 2017, Tokula and Ejaro, 2018, Madallah and Tarawneh, 2014, Jiang et al., 2013). In addition to this, the combination of data gained from satellites with spatial metrics; as suggested by Abebe (2013) and Milad et al. (2017), was used as a quantitative determination of urban expansion and sprawl pattern to understanding and representation of spatiotemporal of urban expansion. Recent information on sprawl dynamics is very crucial to monitor and visualize the growth pattern and provide up-to-date information for urban planning and management (Saxena et al., 2016) as well as for the proper provision of service and infrastructure (Jiang et al., 2013) and (Kii and Nakamura, 2017). In the case of urban sprawl dynamics and growth changes, change caused by rapid and uncontrolled urban expansion has significant negative impacts affecting the environment, ecosystem and society (Haas, 2013, Das and Das, 2019, KHAREL, 2010). These impacts are specifically important for developing nations where there is limited capacity to cope with environmental and social effects of rapid urban expansion.

The uncontrolled and illegal expansion of urbanization (urban sprawl) and its impacts are being studied better in megacities in different parts of the world (Qian and Zhen, 2019, Herold et al., 2003, Sun et al., 2014, Khouri et al., 2018, Andre Sorensen and AndreJunichiro Okata, 2011); but, studies in this area are not well investigated and documented for small and medium cities of economically poor countries because of the absence of reliable and advanced spatial and socio-economic data (Cohen, 2006). Recently, integrated approaches which use the combination of satellite data that have high spatial accuracy and freely available or less expansive satellite images with GIS that allows to quantitatively measure the pattern and rate of the rapidly growing cities and urban areas in a rational and justifiable manner (Dadras et al., 2015, BHATTA, 2009, Milad et al., 2017, Luo, 2008, Li, 2014). Currently, various studies on techniques of remote sensing such as supervised classification (Thapa and Murayama, 2009, Almazroui et al., 2017, Moeller, 2007, Odjugo et al., 2015) and unsupervised classification (El et al., 2017, Mosel et al., 2016, Ahmad and Goparaju, 2016, Peacock, 2014, Nassar et al., 2011, Nong et al., 2018, Jain and Dimri, 2016) were effectively used to monitor urban sprawl and expansion in various spatiotemporal scale. In addition to these, Shannon entropy and spatial metrics were also used to quantify urban sprawl (S, T. N. T. S and Truu, 2017, Deka et al., 2012) and determine the different dimensions of sprawl (Krishnaveni and Anilkumar, 2020, Shenbagaraj and Stalin, 2019). Shannon entropy index is a commonly used technique to verify and identify the disparity of urban sprawl and urban areas (Nkeki, 2018, Alsharif et al., 2015, Das et al., 2016, Herold et al., 2003, Muiruri and Odera, 2018, Cai et al., 2015) while spatial metrics are helpful to quantify and measure the spatial characteristics of the landscape (Dutta and Das, 2019; Ramachandra et al., 2014).

Ethiopia, the second largest country in Africa in terms of population size, has around 116 million population, 4.6% average annual urban growth rate and 2.4% annual growth rate, which is higher than the average growth rate (2.2%) of developing countries (UN, 2016). Recently, Ethiopian economy is experiencing broad base growth (AbafitaAbbas et al., 2018) and economic development and rapidly growing population together led to unplanned urban expansion in different parts of the country (Terfa et al., 2017). In the case of Ethiopia, urban sprawl is manifested in excessive and unlimited outward expansion of urban areas at the expense of other land use classes in a way that contradicts the spatial plan of the city. As the findings of the study conducted by Zemenfes G (2014) in Mekelle city, Ethiopia, the city was sprawling into the nearby rural community as a result of the uncontrolled and unauthorized acquisition and occupation of farmlands; which is mainly related to poor land use and administration policy of the city. Several studies were conducted on urban expansion and urban land use change in Ethiopia with numerous driving forces, which include both proximate and topographic variables and its impact on livelihood and environment (Deribew, 2020, Barow et al., 2019, Fenta et al., 2017a, Fenta et al., 2017b, Fenta et al., 2017c, Kindu et al., 2020, Erasu, 2017, Gebremedhin et al., 2019, Halefom et al., 2018, Gashu and Egziabher, 2018, Fitawok et al., 2020, Terfa et al., 2017). For instance, studies conducted by Deribew (2020) in the western fringe of Addis Abeba city, Ethiopia, Barow et al. (2019) Jigiga town of Ethiopia, Fenta et al., 2017a, Fenta et al., 2017b, Fenta et al., 2017c in Mekelle city of northern Ethiopia, (Halefom et al., 2018) in Debre tabor town of south Gonder zone of Ethiopia and Gashu and Egziabher (2018) in two Ethiopian cities: Bahir Dar and Hawassa, recognized rapid urban expansion towards agricultural areas and forest land. Slope, road accessibility and distance to CBD were considered as a major proximate driving forces of rapid urban expansion (Fitawok et al., 2020).

Attempts have been done to analyze urban expansion and its response to the urban landscape in the study area. For instance, Fitawok et al. (2020) tried to analyze the impact of urbanization and land use change in Bahir Dar and reported that uncontrolled rapid urban expansion occurred at the expense of farmland and vegetation from 1991 to 2019. Haregeweyn et al. (2012) was also conducted a study on spatiotemporal trends of urban land use land cover and green infrastructure change in Bahir Dar and Hawassa from 1973 to 2015. The study depicted that there was a dramatic change in green urban areas due to rapid expansion of built –up areas in the study period from both cities. Study conducted by Admasu et al. (2018) also provide analysis on socio –economic impacts of urbanization on the local land in the periphery of Bahir Dar. The results of this study indicated that rapid urban growth of Bahir Dar from the center to the periphery has a significant socio –economic impacts on the land owned by the surrounding communities. Previous studies assessed spatiotemporal trends of urban land use change and its impact on the surrounding community considering the physical and proximate variables as main driving factors of growth without giving much attention to its drivers. Specially, these studies completely ignored the significant impacts of socio- economic and demographic driving factors on urban growth in Bahir Dar. In addition to this, all prior studies conducted in Bahir Dar have not been used spatial metrics for their studies. So, these studies gave partial conclusion on urban expansion since the complex urban system and its shape, dimension, landscape structure and growth pattern can be determined through these metrics calculation. Even though, these studies were conducted based on satellite data using GIS and RS, metrics were not used to determine and quantify the landscape structure and extent of sprawling. These all were based on land use change detection which is insufficient to give conclusion about the existing complex urban system. Furthermore, Bahir Dar has different spatiotemporal characteristics as compared to other study areas with similar socio-economic conditions such as Jigjiga, Mekelle, Gonder and Hawassa. Though urban growth is a common phenomenon for all cities, Bahir Dar has been sprawling in a fastest rate covering large area within a short period of time and rapid development of new fragmented patches. So, this study has been conducted to address these gaps which were not considered as significant issues, and makes the study different from prior studies in terms of the method employed, study time and data used.

This study was conducted in Bahir Dar city. The city was purposively selected because it is the epicenter of economic, cultural and political activities of the country in general and of the regional state in particular. These activities are aggravating rapid and unplanned urban expansion which in turn leads to urban sprawl in the city. The degree, pattern, rate and spatiotemporal characteristics of the city’s urban sprawl and the consequences on the other LULC categories are not well investigated to the required extent. Especially, the application of Shannon’s entropy and spatial metrics to monitor and analyze urban sprawl which is potentially significant to obtain quantitative information is not used till this study has been conducted in Bahir Dar. Furthermore, this study was conducted by integrating spatial metrics and sociological researches/surveys and results obtained through triangulating these two methods provide better understanding and inferences. So, this study has been conducted to analyze the spatiotemporal dynamics of urban sprawl and monitors the extent of its growth pattern for the last three and half decades (35 years) from 1984 to 2019 in Bahir Dar. Basically, the study was conducted on three interrelated objectives. The first objective was to understand the spatiotemporal dynamics of urban sprawl over Bahir Dar city, Ethiopia. The second objective was to quantify the extent and rate of urban sprawl and its associated driving forces. The last objective was to examine the different dimensions of urban sprawl using spatial metrics. Therefore, it is important for urban planners and policy makers to understand multidimensional phenomena of urban sprawl and its spatiotemporal dynamics in the city which can enable effective utilization of resources and allocation of infrastructure by controlling improper enforcement of land use. Furthermore, the findings of this study are beneficial to reduce the sprawl and its socio –economic and environmental impacts by providing directions to the local development authorities.

2. Materials and methods

2.1. Study area

Bahir Dar city is one of the largest and rising cities of Ethiopia. It is also the major tourist attraction site of the country because of the presence of Lake Tana and cultural heritages such as Lake Tana monasteries and different annual religious festivals. Bahir Dar city is well known for its biodiversity and is registered by UNESCO as one of the best cities in the world in its biosphere reservoirs (Worku, 2017b, Worku, 2017a, Woldu, 2018). This research has been done in Bahir Dar City and its surroundings which is the center of Amhara Regional State. The geographic location of the Bahir Dar extends from latitude 11°30'0'' to 11°41' N longitude 37°16' to 37°30'0'' E (Fig. 1). It is found in the northwest part of Ethiopia at a physical distance of 565 km from Addis Ababa, the capital city of Ethiopia (Fig. 1).

1. Download: Download high-res image (720KB)
2. Download: Download full-size image

Fig. 1. Map of the study area.

The population of Bahir Dar city grew from 54,800, first national census, to 96,140, second national census, with an increase of about 75% from 1984 (CSA, 1991). In the other census which took place in 2007, the total population has become 221,991(CSA, 2008). The value increased by about 87% of the census of 1994 according to the CSA. Census is conducted in every 10 years interval but because of different political and economic reasons, the current population of the city is not exactly known because the census has not been conducted recently. Based on the sample survey conduct in 2017, the total population of Bahir Dar was 348,429.

2.3. Methods of data analysis

2.3.1. Change detection analysis

Post classification comparison integrated with ArcGIS desktop has been used on the classified thematic maps of the study years. This technique has been utilized for change detection due to manageable land use maps, the small number of land uses and has been confirmed to have radical land use change (Yang and Wen, 2011). The change in each class that is built-up, forest, water, and wetland and crop land has been calculated properly. The following formula has been used to calculate the rate of change hectare/year and percentage share of each land use class in the study period (Meshesha et al., 2016):(3)$\mathbf{\Delta }\mathbf{A}\left(\%\right)={\mathrm{A}}_{\mathrm{t}1}-{\mathrm{A}}_{\mathrm{t}2}/{\mathrm{A}}_{\mathrm{t}1}*100$Where $\mathbf{\Delta }$A% = percentage of change in the area of land use land cover class initial time At₁ and recent time period At₂, At₁ = area of land use land cover class at initial time, At₂ = area of LULC type at recent time. As stated by (Fenta et al., 2017a, Fenta et al., 2017b, Fenta et al., 2017c), the change rate of LULC type in ha/year was calculated by the following:(4)$\mathit{R}\mathbf{\Delta }(\frac{\mathrm{ha}}{\mathrm{year}})=\mathrm{Z}-\mathrm{X}/\mathrm{W}$Where R rate of change, Z = recent area of land use land cover type in ha and X = previous area of land cover type in ha and W= time interval between Z and X in years.

2.3.2. Urban sprawl measurement

Shannon entropy method is a technique used to determine the extent of concentration or diversion of built-up areas in urban growth and sprawl study. It is commonly used to determine whether the change in urban areas was divergent or compact based on the Shannon entropy index values (Mashagbah, 2016). Based on the study of Mundhe and Jaybhaye (2015), the study area was divided into 12 zones using concentric circle with consecutive incrementing radius of 1 km radius created around the center of the city (CBD), with center at location St. George church in 11° 35' 43'' N and 37° 23' 20' E'. Shannon entropy index for this study was calculated (Bhatta et al., 2010) by:(5)${\mathrm{E}}_{\mathrm{n}}={\sum }_i^n\mathrm{pi}*\log (\frac{1}{\mathrm{pi}})/\log (\mathrm{n})$Where Pi = X_i/${\sum }_i^n\mathrm{Xi}$ and X_i is the density of land development, which is equal to the amount of built-up land divided by the entire amount of built-up land in the i_th of n total zones (Tewolde and Cabral, 2011, Mohammady and Delavar, 2014). The type of entropy used in this study was relative entropy which enables to calculate the entropy value into a value that ranges from 0 to 1 (Bhatta et al., 2010) and this entropy value indicates the concentration of built-up areas that shows the highest density in one region if the values are zero or close to zero and the value is 1 when the built-up areas have dispersed spatial distribution (Tewolde and Cabral, 2011). If the entropy value crosses the half marks (0.5), the city is experiencing rapid sprawl with unevenly distributed dispersion (Rastogi and Jain, 2018, Singh, 2014). The difference in entropy between time t (initial) and t₁ (final) was used to reveal the magnitude of change in urban sprawl by calculating the difference in disparity by using:ΔE_n=E_n(t+1)–E_n(t) (6)Where ΔE_n is the difference in relative entropy from two time periods, E_n (t + 1) is the relative entropy value at time t + 1 and E_n (t) is the relative entropy at time t(H. Sun et al., 2007).

Spatial metrics were also used in this study. These metrics are tools that provide quantitative information on the level of urban sprawl and understand the complex and dynamic nature of the spatiotemporal pattern of growth of urban areas in quantitative ways (Nong et al., 2018, Reis et al., 2016). Large number of metrics has been developed to quantify urban sprawl at class and landscape levels. But it is important to select appropriate metrics out of the whole metrics available suiting to the objective of the research (Sinha, 2018). There is no one best spatial metrics that can provide all aspects of landscape characteristics; the choice of the metrics ultimately depends on the purpose of the problem under investigation, nature of landscape, easy interpretation and the potential of each metrics to best describe urban sprawl (Riitters et al., 1995, Deng et al., 2009, Charles Dietzel et al., 2005). Based on these literatures, 13 frequently used metrics were initially selected to examine the urban sprawl dynamics on the landscape of Bahir Dar. The outputs of these metrics, which were processed and analyzed using FRAGSTAT V4.2, were statistically evaluated using SPSS. Then, a highly correlated metrics having a coefficient of greater or equal to 80% (Pearson’s correlation coefficient) were excluded for further analysis to remove/minimize redundancy. Therefore, the 6 metrics with the least correlation between each other were chosen for this study. These metrics which were used for this study reveals the different dimensions of urban sprawl such as pattern, shape, density and surface. Spatial landscape metrics which were used in this study were total area (TA), number of patches (NP), largest patch index (LPI), edge density (ED) and area weighted mean patch fractal dimension (FRAC_AM). These metrics were selected properly among the metrics which can be used for land fragmentation analysis to eliminate the duplication of information obtained during analysis. These metrics were performed for the classified LULC maps (built up and non- built up) using FRAGSTAT V4.2.

2.3.3. Metrics calculation and sociological research

Sociological research is the systematic study of individuals, people and social phenomena using measurement techniques such as surveys, interviews, FGD or detailed text analysis (Martin, 2017, Igwe and Odii, 2020). Whereas, spatial metrics are numerical measurements that can be very useful in understanding urban expansion and urban sprawl since they represent the physical dimension of both well (Sinha, 2018). In this study, both of the two methods have been used to obtain the required data and extract the required information that enables the researchers to reach on conclusion. Spatial metrics were used to determine the existing urban sprawl and study the dimension, shape, structure and patterns of landscape of Bahir Dar; which were extracted from land use maps, using FRAGSTAT 4.2. The existing urban sprawl has been determined using these metrics and the driving forces for this phenomenon cannot be identified using these metrics. So, another method called sociological research; research based on the data collected directly from residents of the city through interview, questionnaire and FGD, has been used to identify the driving forces of urban sprawl in the study area. Here, both methods were used and provide necessary information for this research. So, the integration of spatial metrics and sociological researches provides a clear insight and understanding about urban sprawl and its factors in the study area (Fig. 2).

1. Download: Download high-res image (595KB)
2. Download: Download full-size image

Fig. 2. Methodological Flowchart of the Study.

3. Results

3.2. Land use land cover map

The study revealed that the share of built-up has been growing from 1.14% in 1984 to 1.68 in 1994, 8.29% in 2009 and 11.86% in 2019. This causes a rapid increase in existing urban sprawl and the continuing trend of urban growth in Bahir Dar (Fig. 3).

1. Download: Download high-res image (2MB)
2. Download: Download full-size image

Fig. 3. LULC maps of Bahir Dar city and its surroundings for 1984, 1994, 2019 and 2019.

3.5. Urban Sprawl Measurement

As indicated in the figures below (Fig. 4), the built-up area of Bahir Dar and its surroundings has been expanded rapidly from CBD (center of urban growth) towards all directions except the north direction, which Lake Tana occupied. This uncontrolled and illegal growth of built-up surface is leading to the rapid growth of urban sprawl in the landscape of Bahir Dar.

1. Download: Download high-res image (935KB)
2. Download: Download full-size image

Fig. 4. Multi circular buffer zones created by 1 km interval around the city center (CBD) in 1984, 1994, 2009 and 2019.

The spatiotemporal distribution of built-up area over the study period (1984 – 2019) has been examined and represented by the figure below (Fig. 5). This indicated that built-up areas becoming more dispersed in recent years as compared to the previous study period. This density of built-up area can be used to examine dispersion (Fenta et al., 2017a, Fenta et al., 2017b, Fenta et al., 2017c). High density values were observed in the core of the city, while the low density of built-up area was found in the outer rings.

1. Download: Download high-res image (161KB)
2. Download: Download full-size image

Fig. 5. Changes in built-up density over time in individual concentric zones in Bahir Dar city and its surroundings.

The calculated value of Shannon entropy ranges from 0.45 to 0.71. Based on entropy method of urban modeling, there was no urban sprawl/compact structure/ in the year 1984 hence the value (0.45) was less than 0.5. The overall values obtained were 0.45, 0.58, 0.71 and 0.74 for the years 1984, 1994, 2009 and 2019 (Table 7). The log (n) value for this study was 1.079 and this indicates the entropy values of 2009 and 2019 were closer to log (n), which shows high degree of dispersion and urban growth has occurred as spread of urban growth. Relatively lower value of Shannon’s entropy (0.45) occurred in 1984 and higher value of Shannon’s entropy was in 2019 (0.74) (Table 7). This indicated that the built-up area of 1984 was more compact and homogenous in the core (CBD). The increased entropy values indicate increase in built-up area and expansion of urban sprawl with dispersed or scattered distribution. There was moderate/slight increase in the share built-up area and Shannon's entropy value from 1984 to 1994, which reveals that Bahir dar has grown towards dispersed development.

Table 7. Shannon's entropy with the magnitude of urban sprawl change (ΔE_n) in Bahir Dar and its surroundings for the years 1984, 1994, 2009 and 2019.

S. No	Years	Shannon’s Entropy Index	Change in Shanon’s Entropy Index ( ΔEn)
1	1984	0.45	Base Year
2	1994	0.58	0.13
3	2009	0.71	0.13
4	2019	0.74	0.03
Log(n) = 1.079

In addition to Shannon entropy index, landscape metrics were also used to testify the presence/absence of urban sprawl and determine its extent (Table 8). Class area (CA) has been increased from 261 to 2566 during the study period (Table 8). There was a continuous increase in the number of patches (NP) which were 371 in 1984 and increased to 3576 in 2019 (Table 8). Similarly, the patch density (PD) of the study has shown increased trends throughout the study period. In 1984 the PD was 1.74 then increased to 5.05, 6.75 in 2009 and 9.72 in 2019 (Table 8). This indicated that PD increased in all study years indicating that there was fragmented development of built-up areas over the study period.

Table 8. Result of spatial metrics computation for the entire landscape of Bahir Dar.

Year	LU/LC	TA/CA	NP	PD	LPI	ED	FRAC_AM
1984	Built -up	261	371	1.74	12	7.09	1.12
1994	Built -up	396	1077	5.05	20	12.86	1.17
2009	Built -up	1825	2518	6.75	48	40.45	1.19
2019	Built -up	2566	3576	9.72	52	46.77	1.25

The result reveals that the largest patch index (LPI) values were 12 in 1984, 20 in 1994, 48 in 2009 and 52 in 2019 (table 8). This might be due to the join up of small and isolated built–up patches into large single patches and or the emergence of other patches near to/around previously existing patches in the study area. When built-up areas are growing towards urban fringes, small and fragmented patches were connected to each other and with the urban core. Edge density (ED) has been increased from 7.09 in 1984 to 12.86 in 1994 and from 40.45 in 2009 to 46.77 in 2019 showing that there has been rapid urban expansion and development of various fragmented urban patches in the landscape of Bahir Dar. In the case of FRAC_AM (area weighted mean patch fractal dimension), it has been 1.12 in 1984, 1.17 in 1994, 1.19 in 2009 and 1.25 in 2019. The value of FRAC_AM closer to one (1) indicates that built-up areas were composed of patches that had simple geographic shape in the earlier periods of urban expansion but an increased trend of FRAC_AM in Bahir Dar and its surroundings indicates that the study was getting more fragmented and complex over time due to rapid urban expansion.

3.6. Causes of urban sprawl

The participants of this study identified 11 factors that aggravate the nature of urban sprawl in the study area. These driving forces can be categorized as demographic, inner city problems, macro and microeconomic factors, infrastructure and regulatory frame factors, which were recognized as important factors that intensify urban growth and sprawl in Bahir Dar and its surroundings (Table 9). Among these, rapid population growth, land use policy, weak land use policy and planning, poor law enforcement, low price of land, high housing preferences, rural-urban migration and poor governance were the 8 top factors of urban sprawl in the study area for the last three and half decades. Rapid population growth, land price, and rising living standard were macroeconomic variables.

Table 9. Driving forces of urban sprawl in Bahir Dar (% taken from Questionnaire participants).

Deriving forces of urban sprawl	% of respondents
	Yes	No
Land use policy	98.6%	1.4%
Rapid population growth	99.8%	0.2%
Poverty and unemployment	74.3%	25.3%
Rapid expansion of infrastructure	55%	45%
Rural-urban migration	75%	25%
Poor governance	78%	22%
Low price of land from farmers of the surrounding area	89%	11%
Unsafe environment from the center of the city	5.4%	94.6%
High housing preference	85%	15%
Weak land use planning and poor enforcement	97.8%	2.2%
Rising living standard	52%	48%

The population of Bahir Dar city and its surroundings has been increasing at an alarming rate and was highly correlated (Fig. 6) with built-up area expansion in the study period (1984 – 2019). As indicated in the figure below, the correlation coefficient (R²) between population and built-up area was R² = 0.96, which means that these two variables were strongly associated with each other and both population and built-up area were shown alarming increment in the study period. The other driving forces such as weak land use planning, policy problems and poor enforcement of land use plan were factors related to the regulatory framework of the city. The result indicated, 99.8% of the respondents perceived that rapid urban expansion and urban sprawl was due to the rapid increment of the population in the city and surrounding areas.

1. Download: Download high-res image (135KB)
2. Download: Download full-size image

Fig. 6. Correlation between population and built-up area.

The FGD participants also proved that rapid urban growth and the existing urban sprawl of the city was the result of rapid population growth, poor urban planning, land use policy, poor governance and low price of land in the surrounding area of the city were the top deriving forces. Policy of the country’s land use, which is also under utilization in the city, encourages the rapid expansion of the city towards the peripheral area and increases the sphere of influences of the city via providing land for each dweller of the city based on their preference. Furthermore, the FGD discussants also noted that the land owned by the surrounding farmers is easily affordable for the poor urban dwellers hence it is available with low price. In addition to this, the urban planners were not that much well educated in the area to manage the city and achieve sustainable urban development. The participates also stated that lack of enforcement of regulatory frameworks related to urban planning enforced the urban dwellers to purchase land from farmers in agricultural land with low price and encourage illegal acts regarding the construction. Besides this, poor governance like the provision of land in a corrupted way also becomes a common maladministration problem, which neglects equality of all dwellers in sharing the resource. The other factors such as poverty and unemployment, rising living standards, rural-urban migration and high preference of housing were also mentioned as important driving forces of fast urban growth and urban sprawl in Bahir Dar.

4. Discussion

Urban sprawl and uncontrolled urban expansion is a pressing problem and global concern throughout the world. It has been increasingly associated with a range of social and environmental problems such as inefficient energy and land use, traffic congestion, environmental degradation, social segregation and isolation (M B Sridhar et al. (2020) & (Hamad, 2020). In this study, we presented an accurate analysis of spatiotemporal dynamics of urban sprawl and urban growth pattern in Bahir Dar over 35 years. Shannon’s entropy and six selected spatial metrics were used to monitor and analyze the spatiotemporal dynamics of the study area. Urban sprawl prevention and sustainable urbanization can only be achieved by properly determining urban sprawl and understanding encouraging forces ( Li, 2012). Every city has a unique and specified morphology. These different morphologies were shaped by physical, demographic and socio-economic factors. Bahir Dar, as one of the oldest cities of Ethiopia, has remained under rapid urbanization processes and has undergone intense anthropogenic changes and transformations (Fitawok et al., 2020). The results of this study indicated that the urban area in Bahir Dar has undergone considerable sprawl. The rapid population growth and the proportional increase in built-up area has been resulting urban sprawl in Bahir Dar. Zhang and Xie (2019) indicated that urban sprawl is mainly attributed to rapid population growth which in turn results in high demand for housing in urban fringes. In addition to the existing population growth, land use policy, weak land use policy and planning, poor law enforcement, low price of land, high housing preferences, rural – urban migration and poor governance were the main deriving forces of urban sprawl as proven by the participants of this study. As studied by (Christiansen and Loftsgarden, 2011), urban sprawl is a common problem in Europe and land price, market failure, high housing demand, rapid population growth, purchasing power, infrastructure development, regulatory frameworks and governance were identified as the drivers of urban sprawl in selected cities of Europe.

Shannon’s entropy values (0.45 – 0.71) showed sprawling characteristics similar to Qom city, Iran (0.80 – 0.88) determined by Mosammam et al. (2017), Tripoli, Libya (0.74 – 0.90) determined by Alsharif et al. (2015) and Chennai, India (0.67 – 0.69) determined by Deka et al. (2012). An increase in the values of Shannon entropy index indicates there was a continuous increase in urban sprawl which tend be more dispersed. Here, change in Shannon entropy index from 1984 to 2019 was 0.16 unlike the previous studies which have an entropy value of 0.08, 0.16, 0.02 and 0.02, respectively. This indicates that the rate of sprawling and urban development of Bahir Dar city is very high in the study period. So, the Shannon entropy values here determine the presence of continuous urban sprawl from 1984 to 2019 in Bahir Dar, and depict the need for safe plans and strategies to manage the city for its sustainable development. Shannon entropy has a practical advantages such as requiring minimum data (land use maps) and easy to calculate using basic software. But it has also disadvantages like sensitivity to the zoning scheme and being affected by the area. It only considers the relative distribution of built –up areas instead of absolute amount. These are the weakness of Shannon’s entropy as compared to other techniques such as Batty’s spatial entropy and grid based zoning scheme which are potentially suggested to avoid such biases for urban sprawl study.

The NP used to indicate the aggregation or disaggregation in the landscape of Bahir Dar city, and the study indicated that there were numerous discontinuous patches of built-up area that occurred as a result of rapid urbanization. The NP patches were increased from 371 in 1984 to 3576 in 2019 between 1984 & 2019 (Table 8). This is similar to the study in Tshwane city, South Africa where there was increments in the NP from 40, 253 in 1984, and 44, 848 in 2015 (Magidi and Ahmed, 2019). As a result of a continuous increase in human activities, the NP patches for all land use classes of the study area were continuously increased and caused landscape fragmentation and irregularities (Wan et al., 2019). Area weighted mean patch fractal dimension (FRAC_ AM) is used to determine the complexity of urban patches and degree of irregularity of urban land use change. The perimeter increased in path area was 1.12 in 1984 and increased to 1.25 in 2019 (Table 8). Similar result was also found in Istanbul metropolitan city where 12.7% increase in FRAC_AM was observed between 2000 and 2009 (Kowe et al., 2015). Landscape metrics helped to determine the spatiotemporal dynamics of urban sprawl and changes in landscape characteristics at different spatial scales. Other landscape metrics such as CA, PD, LPI and ED in Table 8 indicated an increase in impervious surfaces in the study period (1984 – 2019). Here, the values of spatial metrics were increased during the study period and this indicates that the urban landscape of Bahir Dar was highly fragmented due to the development new patches in different parts of the study area over the study period. Furthermore, these values proven that there was spatiotemporal variation and dynamics of urban sprawl in the process of the city’s development hence the values vary over time. The changes in values of spatial metrics were also very high as compared to other prior studies though the study varies in terms of study period and total area. High changes of metric values in small area were recorded in Bahir Dar as compared to studies conducted in large metropolitan cities which confirmed presence of rapid urban sprawl. The results of the study conducted in Jaipur district of India by (Dadhich and Goyal, 2017) indicated that the NP, ED and LPI were decreased in its values which means that the development of new patches and the increments of seize of patches was very low. This may be due to low urban growth rate as compared to Bahir Dar city and others. Urban growth is very complex phenomenon but these metrics makes it easy to quantify and measure the spatiotemporal dynamics and its pattern easily. Based on facts established in this study, built-up areas shown continued increase at the expense of non-built land. In addition to the socio –economic and demographic drivers, this landscape fragmentation and dispersed growth in Bahir Dar may be caused by other factors such as absence of physical constraints. In Bahir Dar, there are no physical constraints such as high elevation and steep slope, which encourages rapid urban growth, development of new fragmented patches and dispersed growth. The developments in Bahir Dar and the findings of this study generally support the diffusion-coalescence theory of urban growth (He et al., 2017, Dietzel et al., 2005, Fang and Zhao, 2018). Though spatial metrics are an effective numerical measure to determine and quantify the spatiotemporal pattern of urban growth, it has also drawbacks in theoretical and technical aspects. For instance, it is constrained by proper theoretical understanding of metric behavior. This problem can lead to erroneous interpretation and incorrect conclusion (Jaeger, 2000). Furthermore, spatial metrics are also challenged by lack of proper reference framework, problem of defining relevant landscape and metrics redundancy (Cushman et al., 2008, Wimberly et al., 2000).

Studies in other cities of Ethiopia with similar socio –economic condition has different level of urbanization and effects of urban sprawl. For instance, study conducted by Barow et al. (2018) in Jigjiga city, Ethiopia from 1985 − 2015 indicated that forest land was increased from 1.4% to 9.9%, crop land decline from 54% to 33% and built –up 2.2–9.9%. This indicates that as urban growth increases, forest also increase but crop land decreases. Studies in other cities with similar socio –economic condition such as Dessie (Belete, 2017), Gonder (Belay, 2014), Mekele (Fenta et al., 2017a, Fenta et al., 2017b, Fenta et al., 2017c) and Hawassa (Terfa et al., 2017) indicated that there is urban sprawl which is affecting the socio economic and environmental aspect of the cities and its surroundings but the degree of consequences and the types of land uses affected are different. This indicates degree and extent of consequences of urban sprawl vary with time and space and impacts of urban sprawl in Bahir Dar were severing because of its fasts growth towards agricultural, forest and wetlands. Furthermore, Bahir Dar has unique spatiotemporal characteristics as compared to these studies which show very fast growth; rapid sprawling within a short period of time and its total area has been increasing in alarming rate.

The study used the combination of landscape metrics and sociological researches/surveys to determine the spatiotemporal dynamics of urban sprawl. This provides better understanding of the problems and the complex urban phenomenon through triangulating results obtained from landscape metrics with sociological researches. Landscape metrics provide quantitative information about the problem investigated in this study while sociological researches gave information by which researchers were not obtained by using spatial metrics about urban sprawl and its deriving forces obtained directly from residents.

5. Conclusion

Investigation and monitoring of spatiotemporal urban sprawl dynamics and spatial patterns of urban growth is a baseline to set up proper and effective urban sprawl and growth policies which helps to maintain sustainable urban development for the present as well as future generation. In this study, multi-temporal satellite images for the year 1984, 1994, 2009 and 2019 were used to prepare the LULC maps, analyze urban sprawl dynamics and measure patterns of urban growth of Bahir Dar and its surroundings. In the last there and half decades (35 years), built-up area increased by 937 ha and mostly gain from crop land. In the present study, an increasing trend of Shannon’s entropy index was observed for the past three and half decades (0.45 in 1984–0.74 in 2019), which revealed the presence of urban sprawl in the study area, Bahir Dar city and its surroundings. The results of spatial metrics also revealed heterogeneity and dispersion of urban growth; which in turn results in urban sprawl, increases gradually from the growth pole (center of the city) towards the surrounding area during the study period. This study provides accurate and timely information to stakeholders include policy makers, urban planners and environmentalists for better understanding of the past, present and future status of urban growth. Furthermore, this study proves that integrated approaches of RS, GIS and spatial metrics including Shannon’s entropy index provides a powerful and effective means to reveal the actual changes in LULC, degree of urban sprawl and spatial patterns of urban expansion through quantifying the remotely sensed data using GIS and FRAGSTAT technologies. The study also tried to investigate the major deriving factors that intensify urban expansion and its sprawl through interview and FGD. Rapid population growth, land use policy, weak land use plan and poor law enforcement, low price of land in the surrounding community, rapid infrastructure expansion and high demand of housing were identified as important driving forces.

Ethics approval and consent to participate

Not applicable.